The Study Of Blind Equalization Algorithm Based On Signal Detection Theory

Blind equalization is a new adaptive equalization technology without resorting to a training sequence, instead the known statistical properties and prior information of transmitted signals are exploited to equalize the channel character, it makes the output sequence of equalizer approximate input signal sequence. It includes to widely contents. In the paper, the blind equalization based on signal detection theory is studied. Its content is rich, from principle, it can divide into four aspects such as maximum likelihood sequence estimation, minimum error probability estimation, Bayesian estimation and maximum a posterior probability estimation.The major contribution of this paper is summarized as follow:1) This paper simply analysis the basic theory of blind equalization, and systematically sums up the basic theory on signal detection, such as decision criterion , parameter estimation and optimal receive with intersymbol interference.2) Particularly study the blind equalization based on minimum errorprobability. In the paper, amelioration is advanced, it is that error probability is educed by iterative way and the weight-value iterative adopts LMS-Newton gradient algorithm based on the cost function which is advanced by Li Daoben. The simulation results show that the new algorithm has higher convergence speed and less steady residual error than conditional algorithm.3 ) Thoroughly analyses the blind equalization based on maximum likelihood sequence estimation. Now, conditional algorithms have the defects of calculation complexity and realization difficulty, in the paper, the algorithm of maximum likelihood sequence estimation based on grouping realization is advanced, it utilizes the grouping way, then Viterbi algorithm process serial calculation, and the calculation complexity turns exponential increasing into linear increasing. Otherwise, the algorithm adopts the way of soft-decision, then the signal output is the format of maximum-likelihood, it reduces signal error transmitting. The simulation results show that the new algorithm has better performance, less steady residual error and lower calculation complexity than conditional ones.